|
GSEA法挖掘TCGA免疫浸润机制，大家学习可以参考这个文章: Cell Reports的文章Pan-cancer Immunogenomic Analyses Reveal Genotype-Immunophenotype Relationships and Predictors of Response to Checkpoint Blockade。 由于时间有限，小伍没有把全部的写出来，分流流程我大致看了一下，差不多可以理解并分析。 这篇文章于2017年发表，利用了GSEA和去卷积等方法。 整个分析策略： 1，从TCGA数据挖掘以下内容： 1）整理出TIL,利用GSEA法和卷积法。 2）包含新抗原和 3）肿瘤异质性; （4）免疫表型CGA Figure 1. Strategy for Pan-cancer Immunogenomic Analyses (A) The scheme shows the immunogenomic analyses and the types of data used for the analyses. The results are deposited in a web-accessible database, The Cancer Immunome Atlas (TCIA) (https:///). (B) Immune-related signatures are derived from expression profiles of purified immune cells, normal cells, and cancer cell lines, and used for the gene set enrichment analysis (GSEA) of the TCGA RNA-sequencing data. 2.免疫浸润的细胞表征揭示了不同癌症中的预后细胞类型
Figure 2. Cellular Characterization of Immune Infiltrates in Solid Cancers (A) Immune subpopulations across 19 solid cancers. Cancers are sorted according to the mutational loadand the immune cell subpopulations according to adaptive and innate immunity, respectively. Lower panel shows results from ssGSEA as bubble plot, where the size of the circles gives the percentage of patients with NES > 0 and q value (FDR) < 0.1,="" and="" the="" color="" indicates="" the="" good="" (blue;="">< 1)="" or="" bad="" (red;="" hr=""> 1) outcome (overall survival). The border indicates adjusted log rank p <> (B) Volcano plots for the enrichment (blue) and depletion (yellow) of immune cell types across cancers for tumors with high, intermediate, and low mutational load calculated based on the NES score from the GSEA. (C) Fraction of samples in which selected immune subpopulations were enriched in all cancers with high, intermediate, and low mutational load. (D) Visualization of the immune infiltrates (averaged normalized enrichment score [NES]) in lung adenocarcinoma (LUAD) using two-dimensional coordinates from multidimensional scaling (MDS) (left panel) and for individual patients and selected cell types based on two-dimensional coordinates from t-distributed stochastic neighbor embedding (t-SNE) (right panel). (E) Volcano plots for the enrichment (blue) and depletion (yellow) of immune cell types across cancers for tumor stage I to IV calculated based on the NES score from the GSEA. 突变负荷和组织背景确定免疫浸润的细胞组成,CGA与CD4 +和CD8 + T细胞的浸润相关. Figure 3. Antigenomes in Solid Cancers (A) Cancer-germline antigens (CGAs) that are associated with CD8+ and/or CD4+ T cells. Blue squares show significant association with the corresponding cancer type (Spearman rank correlation > 0.3; adjusted p < 0.1).="" cgas="" are="" sorted="" according="" to="" the="" number="" of="" cancers="" with="" significant=""> (B) CGA expression score across cancers. Cancers are sorted according to the mutational load. (C) Neoantigen load across cancers. (D) Neoantigen frequencies for solid cancers. (E) Neoantigen load for different tumor stages (Kruskal-Wallis test followed by two-sided Dunn’s pairwise post hoc tests on rank sums with Benjamini-Hochberg adjustment of p values). (F) Fractions of neoantigens and their origin. (G) Shared neoantigens in solid tumors. Shown are only neoantigens shared in at least 5% of the tumors. 肿瘤的基因型决定免疫表型和肿瘤逃逸机制 Figure 4. Genotypes and Immunophenotypes in Solid Cancers (A) Mutational load and tumor heterogeneity (two-sided Wilcoxon rank sum test). (B) Immune infiltrates in tumors. Shown is a volcano plot for tumors with high and low heterogeneity calculated based on the NES score from the GSEA. (C) Mutational load and neoantigen frequency (two-sided Wilcoxon rank sum test). (D) Immune infiltrates in tumors. Shown is a volcano plot for tumors with high and low antigenicitycalculated based on the NES score from the GSEA. (E) Hierarchical clustering of immune cell composition for BRAF- and RAS-mutated THCA tumors. (F) Volcano plot for BRAF- and RAS-mutated TCHA tumors calculated based on the NES score from the GSEA. (G) Gene ontology (GO) analysis of the differentially expressed genes for BRAF- and RAS-mutated TCHA tumors using ClueGO (Bindea et al., 2009). (H) Expression of MHC and immunomodulatory molecules in BRAF- and RAS-mutated TCHA tumors. Expression values were compared with normal tissue (log2-fold changes are color coded according to the legend). (I) Volcano plots for SKCM genotypes calculated based on the NES score from the GSEA. (J) Mutational load for SKCM genotypes (Kruskal-Wallis test followed by two-sided Dunn’s pairwise post hoc tests on rank sums with Benjamini-Hochberg adjustment of p values). (K) Expression of MHC and immunomodulatory molecules for SKCM genotypes. Expression values are represented by Z score calculated across all SKCM tumors and color coded according to the legend. 机器学习识别固体癌症中肿瘤免疫原性的主要决定因素 疫表型和对检查点阻滞的反应 |
|
|
来自： 生物_医药_科研 > 《TCGA-肿瘤免疫浸润》
